一种基于投影稀疏表示的基因选择方法  

A gene selection method based on a sparse representation projection

作　　者：耿耀君[1] 张军英[1] 

机构地区：[1]西安电子科技大学计算机学院,陕西西安710071

出　　处：《哈尔滨工程大学学报》2011年第8期1024-1028,共5页Journal of Harbin Engineering University

基　　金：国家自然科学基金资助项目(61070137);国家自然科学基金重点资助项目(60933009);陕西省科技攻关资助项目(2009K01-56)

摘　　要：现有基于稀疏表示的基因选择方法通过回归类标的值确定基因的相关性.对于同一个问题,类标的取值不同,对应的基因选择结果也不同.针对这一问题,提出了一种基于稀疏表示基因表达数据一维投影的基因选择方法.该方法用稀疏表示技术回归基因表达数据在其最可分的方向上的投影,并根据回归得到的基因集的相关性向量选择基因.在6组基因表达数据上的实验结果表明:所提方法的运行时间适中,选择的基因集识别率高的同时冗余度也比较低.Existing gene selection methods based on sparse representation determine the relevance of a gene by regressing the value of a class label.For the same problem,the class label takes different values;the corresponding gene selection result is also different.To solve this problem,a gene selection method based on sparse representation of a one dimensional projection of microarray gene expression data was proposed.The presented method used a sparse representation technique to regress the projection in the most separable direction of microarray gene expression data,selecting genes according to the relevance vector which was obtained by regression for the gene set.Experimental results on six microarray gene expression data sets show that the running time of the presented method is suitable for gene selection and the gene set selected by the proposed method has not only a high recognition rate but also a low redundancy.

关 键 词：基因选择 稀疏表示 基因表达数据 间隔最大化判别分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]